Digital grids need social science as much as engineering, researchers argue

Electricity grids were built as centralized machines for moving power from large fossil-fueled plants to passive consumers, and that design is now colliding with a renewable future defined by thousands of scattered, weather-dependent generators. In a Forum article published in Energy, Sustainability and Society on 23 September 2026, researchers from Germany argue that this collision cannot be solved by engineering alone. Digitalization and decentralization, they write, are the key factors in the transition from fossil fuels to renewable energy sources, and understanding them requires treating the grid as a socio-technical system in which technologies, actors, rules and institutions all change together.

The scale of the shift is measurable. Worldwide, the share of solar and wind power in the electricity production mix rose from 3.8 percent in 2014 to 15 percent in 2024, and renewables reached 32 percent of global electricity generation in 2024, up from 18 percent in 2000. The authors note that the technical and economic potential of renewable electricity production exceeds current and near-future demand, so the question is no longer whether renewables can cover demand but how to achieve efficient upscaling with existing and new technologies. Yet energy-related greenhouse gas emissions remain the largest and fastest-growing driver of atmospheric emissions growth, and electricity generation offers far greater potential for rapid decarbonization than industry, building or transportation.

The article, written by and for social, computer and engineering scientists, is largely focused on developments in Germany, where changes are informed by shifts in the broader European energy market. The authors set out a research agenda for studying the digitalization of electrical grids in an interdisciplinary setting, examining the relationship between grids and society, technical developments in electricity production, and the principles of programming grid management. Their central claim is that analyzing the transition requires a socio-technical understanding: technical research must factor in the socio-cultural logics that electricity systems developed over more than a century and a half, while social research must be grounded in the technical materialities of existing grid infrastructures.

Grids, they argue, are not mere reflections of technological possibility. Their specific forms are informed by history, law, commerce and values, making them large socio-technical systems rather than purely technical ones. Through the liberalization of energy markets in the EU, including the unbundling of electricity producers and grid operators, and the emergence of direct marketers and energy traders, the network that used to distribute electricity has become a field for arbitrage profiteers, even though the efficiency of such solutions for reducing greenhouse gas emissions is highly contested. Decentralization itself is defined in the article as the respatialization of electrical generation capacity in light of the need to decarbonize electricity production, a change the authors link to digitalization as a twin transition.

The researchers identify barriers to decentralized energy production in four domains. Socially, some communities heavily implement the energy transition while others protest against new wind or solar parks, and fossil fuel sector lobbying at international, national and sub-national levels contributes to slowing transitions. Technically, the volatility of renewable sources challenges a grid infrastructure that requires a constant level of frequency and load. Economically, changing subsidies, electricity market schemes and energy pricing mechanisms challenge investments that rely on calculable rates of return, while regulatory regimes fostering ever larger solar parks and wind turbines have turned distributed energy resource installation into an investment option rather than a citizen energy option. Politically, the authors see a lack of coherent climate change mitigation action at international and national levels, alongside legal fights over regional planning and the placement of specific installations.

Digitalization adds its own tensions. The increasing integration of information and communication technologies into grid infrastructures promises new efficiencies and participatory models, but it also creates new problems and vulnerabilities, including digital inequalities and cybersecurity risks. Empirical studies indicate that engagement with smart-grid technologies correlates with income, education and digital literacy, suggesting uneven capacities among consumers to benefit from emerging energy services. The growing interconnection of supervisory control and data acquisition systems has expanded the attack surface of modern energy infrastructures, with cyberattacks through malware such as Industroyer, Triton and NotPetya illustrating vulnerabilities that can compromise system stability and data integrity. Recent research therefore advocates embedding cybersecurity and data-protection principles, from network segmentation and encryption to participatory data governance.

On the technical side, the article traces the advances that made the transition possible. Wind turbine capacity improved to between 12 and 15 MW in the early 2000s, with the horizontal-axis design dominant because it can regulate rotor speed, manage power output and use blade-pitch regulation to prevent damage in severe wind. Solar cells have progressed through thin-film second-generation designs toward third-generation concepts, with laboratory efficiencies reaching up to 46 percent, though about 90 percent of solar cells produced today remain first-generation crystalline silicon due to cost advantages and technological maturity. Battery storage capacity in Germany increased from 2.34 GWh in 2020 to 23.03 GWh in 2025, with available power rising from 1.58 GW to 15.52 GW, and lithium-ion batteries make up 98 percent of German home storage systems and 95 percent of industrial ones.

Yet a fundamental constraint persists: despite these improvements, a far greater number of wind turbines and solar modules are required to generate the same amount of electricity as fossil-fueled power stations. Larger wind turbines have a capacity of about 10 MW and large solar parks might reach 500 MW peak capacity or more, similar to natural gas plants, while contemporary coal-fired plants range roughly between 1,500 and 3,000 MW. Renewable generation therefore retains decentralization as a basic feature, and distributed energy resources must be aggregated and managed through automated or digitalized systems that constantly balance electricity input and output to guarantee constant frequency and voltage. Germany has made comparatively little advancement on the demand side, with a smart meter rollout of around 14 percent in 2024.

A standard form of smart grid technology is the virtual power plant, a combination of software and hardware that aggregates distributed energy resources into a single portfolio, allowing them to be steered and operated in the grid and market as a single power generation unit. Virtual power plants are essentially operating digital twins of heterogeneous resources, and their algorithms optimize energy production, storage and transmission. From a legal perspective in the European Union they are not recognized as a distinct regulatory category but function within the Clean Energy Package, operating mainly through independent aggregators under the Electricity Directive (EU) 2019/944. Their primary objective is often cost-effectiveness rather than grid stability, typically expressed as a mixed-integer linear programming formulation, while the EUPHEMIA algorithm officially couples day-ahead power markets across European price coupling regions, performing hourly calculations that will be reduced to 15 minutes in the future.

The interdisciplinary research agenda emerges from a project called Energy – Digital and Decentral, whose preliminary findings span all three disciplines. Computer scientists are developing a self-adaptive data-integration framework based on the MAPE-K loop, whose prototype monitors simulated HTML data sources, detects structural and content changes with high recall, and achieves about 32 percent accuracy across 25 change types, well above random guessing at roughly 4 percent but still limited, with automated adaptation not yet realized. Process engineers evaluated a dozen machine learning algorithms for forecasting power load and production and identified recurrent neural networks and long short-term memory models as most suitable, with the Time-LSTM variant seen as likely to perform best because it handles long-term dependencies and irregular time steps. Social scientists, using narrative and semi-structured interviews and participant observation with German grid operators and plant operators, found that while production units decentralize, data centralizes as a result of digitalizing the grid, that engineers are pressured to acquire data science competences, and that workers insist on a red line in automating critical infrastructure: as one interviewee put it, AI and automation are just counsellors that make suggestions, and the human must always make the decision in the end.

Subject of Research: Interdisciplinary socio-technical research agenda for digitalization and decentralization of electricity systems in the renewable energy transition

Article Title: A socio-technical approach to digitalization and decentralization in the transition of energy systems

Article References: Müller, K., Karol, S., Ortwein, A., Anyadike-Danes, C. M., Möhring, M., Ahmed, S., & Schumacher, K. (2026). A socio-technical approach to digitalization and decentralization in the transition of energy systems. Energy, Sustainability and Society, 16(1), Article 48. https://doi.org/10.1186/s13705-026-00607-6

Image Credits: AI Generated

DOI: 10.1186/s13705-026-00607-6

Keywords: electricity grids, digitalization, decentralization, renewable energy, virtual power plants, socio-technical systems, smart grids, energy transition, Germany, cybersecurity, machine learning, distributed energy resources